Buoyancy Control of a Semiautonomous Underwater Vehicle for Environmental
Monitoring in Baltic Sea
Madis Listak
Maarja Kruusmaa
Department of Computer Engineering
Tallinn University of Technology
Ehitajate tee 5, 19086 Tallinn
ESTONIA
[email protected]
Institute of Technology
Tartu University
Vanemuise 21, 51014 Tartu
ESTONIA
[email protected]
Abstract – This paper describes a preliminary prototype of a
semiautonomous underwater vehicle. The vehicle is designed for
environmental inspection in a Baltic Sea region. The
environmental characteristics and the purpose of the vehicle set
several restrictions to the vehicle’s design. The concept
represented here aims at meeting these restrictions. The paper
focuses on describing a novel buoyancy control mechanism
based on two controllable lateral ballast tanks. The buoyancy
control permits using the vehicle in two modes - horizontally
compressed and vertically compressed. These modes are used in
different environmental conditions and for different monitoring
tasks.
I. INTRODUCTION
Our goal is to design a vehicle for environmental
monitoring in Baltic Sea that is in the first place meant for
monitoring of underwater vegetation and benthic
morphology.
Baltic Sea is one of the most severely polluted seas in the
world. The extent and distribution of underwater vegetation
gives a lot of information about pollution, climatic
conditions, ice conditions etc., therefore vegetation is
monitored regularly. At present underwater monitoring is
done by divers, which is laborious, expensive and dangerous.
We have designed a prototype of a vehicle that is equipped
with an underwater camera and is meant to replace the
human diver. In addition to hydrobiological surveys the
vehicle can also be used to record other environmental
parameters like water temperature, pH-level, salt content, etc.
Since the device is equipped with an underwater camera it
can also be used for underwater inspection e.g. at rescue
operations, construction work, etc. in shallow waters.
The paper at hand describes the preliminary prototype of
the vehicle and focuses on a novel buoyancy control
mechanism. The buoyancy control permits using the vehicle
at different orientations, either horizontally or vertically
compressed, depending on the environmental conditions and
the task specification.
This paper is organized as follows. In the rest of the
introductory part we describe the task requirements. Next, we
describe our buoyancy control mechanism. Sections 3, 4 and
5 describe the mechanical, pneumatic and control system of
the prototype respectively. We finish this paper with
concluding remarks and an outline for the future work.
A. Task description
The requirements determining our vehicle’s design can be
divided into two large categories. The first category consists
of environmental factors that are unique to the Baltic Sea.
The second category is the human and task specific factors.
In the following subsections we describe both of the
categories closer.
B. Environmental factors
Baltic Sea is different in several senses from subtropical
and tropical seas or exposed seas like Nordic Sea. The factors
influencing our choices for the vehicle’s design are the
following:
1) Depth: Baltic Sea is relatively shallow and therefore the
underwater vehicles do not have to operate under high
pressure.
2) Turbidity: water is very turbid due to floating detritus
and therefore visibility is very low.
3) Extension of vegetation: due to low visibility, the
euphotic zone (the zone of penetration of light into the water
column) is rather narrow and therefore the vegetation is
limited to the costal regions in the depth down to
approximately 20m. The regions of interest for
environmental monitoring are usually near the coastline in
the depth from 1m to 6m.
4) Properties of the seabed: most of the seabed in the
depth of interest is covered with mud or small particles of
zoo- and phytoplankton that has settled down and is
extremely volatile.
5) Human inhabitancy: Baltic Sea is under a severe
anthrophic pressure. Costal regions that are especially
interesting for monitoring are full of harbours, beaches,
fishing nets, dense surface traffic, etc.
C. Human- and task specific factors
This research is strongly demand-driven and therefore we
pay lot of attention to satisfying the requirements of the
users, which are in the first place environmental scientists.
The human and task specific factors that influence our
vehicle’s design are the following:
1) The main function: the main function of the device is to
facilitate environmental monitoring. The highest priority is to
facilitate monitoring of underwater vegetation.
2) Additional functions: if possible, the device should also
permit measuring other environmental parameters, like
temperature, pressure, salt content and make it possible to
correlate these parameters with hydrobiological data. It could
also be used for other type of benthic surveys and diving
under ice in winter.
3) Operational requirements: the device should be
portable, preferably operated by one person only; it should
support storing, processing and mapping of environmental
data.
4) Cost requirements: since the vehicles can occasionally
be lost, their cost has to be kept as low as possible.
preliminary calculations, replacing the diver with a vehicle
like this will be approximately 10 times more efficient.
2) Remote control mode: the operator can manually drive
the vehicle. This mode requires more control surfaces for
greater manoeuvrability. This mode can be used for closer
inspection of underwater sites or diving under ice.
II. DESIGN CONCEPT
B. Fin-like control surfaces
Categorization provided in [1] divides underwater vehicles
according to their purpose into three categories. The first
category is commercial vehicles used mainly by offshore oil
and gas industry for exploitation and exploration of oil and
gas fields. These vehicles are usually heavy, they tolerate
high pressure that makes them applicable in deep ocean
surveys and they are very expensive. Military vehicles are
basically used for reconnaissance, intelligence gathering and
demining, they often operate in a fleet, are fully autonomous
and expensive.
The third category is low cost academic research tools, like
for example [2]. The vehicle described in this paper certainly
falls into the last category. Its main function is to support
scientific benthic surveys. Also its cost has to be kept low to
afford using several vehicles along the costal line and to
decrease the risk of loosing some of them in fishing nets or
on underwater rocks.
According to the restrictions specified above we describe
the main features of the design.
A. Semiautonomous and remotely operated modes
Underwater navigation and mapping is still a field of
intensive research. Despite of several emerging solution, the
methods of underwater navigation are still under
development [3]. In addition, a fully autonomous vehicle
must carry its own batteries. The batteries have to be charged
often and they increase the weight of the vehicle. A fully
autonomous vehicle is more likely to be lost (environmental
researchers report every year loss of theft of a great deal their
equipment). Considering these disadvantages we propose a
semiautonomous vehicle that can be operated in two different
modes:
Fig. 1. Using the vehicle in the towing mode.
1) Towing mode: in this mode the vehicle is towed behind
a boat or a ship (see Fig. 1). With the help of the bow sonar it
adjusts its height from the bottom. The power supply and
localization unit are on the surface as well as the data storage
for the gathered data. The towing mode permits covering
large distances at speed. At present, benthic surveys are often
done by towing a diver behind a boat. According to
Underwater vehicles almost exclusively use thrusters for
locomotion. This is an effective means of propulsion but as a
side-effect it generates high turbulence. The bottom of Baltic
Sea is to a great extent covered with very lightweight
detritus, specially the regions of the greatest interest for
environmental monitoring. Tests with underwater vehicles
show that at the moment the thrusters are switched on,
visibility becomes practically zero and it takes a long time
before the extremely volatile detritus settles down again. We
therefore have decided to use elevators instead of the
thrusters in a towing mode to control the depth of the vehicle
and additional rudders for yaw stability.
In the future, we aim at using the caudal-fin like
propulsion in the remotely controlled mode and pectoral-fin
like motion for stability and manoeuvrability [4][5], but this
is still a topic for future research.
C. Orientation of the vehicle
The novel aspect of this vehicle is a flattened streamlined
body that can be used at different orientations, either
horizontally or vertically compressed (Fig. 2).
Fig. 2. Orientation of the vehicle: vertically compressed (to the left)
and horizontally compressed (to the right).
The two orientations are used in different conditions.
1) Horizontally compressed. This orientation will be
mostly used in the towing mode. The advantage of the
horizontally compressed shape is that the whole body of the
vehicle operates as an elevator and works against lift forces.
Horizontally flattened body can also be used to submerge to
the seabed for a closer inspection. Since its cross section is
smaller than of the vertically compressed body, it is easier to
stabilize the vehicle in lateral currents.
1) Vertically compressed. This orientation will be mostly
used in the remote controlled mode and in a towing mode at
low speeds in very shallow waters with opulent vegetation.
Costal regions interesting for hydrobiologists and other
environmental scientists are often shallow bays fully covered
with underwater vegetation. Some algae can grow up to 1m –
1.5m and reach up to the water surface. A vehicle with a
vertically compressed body is much less likely to get stuck in
the dense vegetation or between underwater rocks. A
laterally compressed vehicle is able to heave and submerge
faster at low speeds and has better manoeuvrability. When a
caudal fin is added, it can be used to propel the body forward
in a remotely operated mode.
The usage of two orientation modes is supported by
biological evidence. Dorsoventrally compressed fishes (like
rays and stakes) are ground-dwelling fishes. Their body
shape is adapted to bottom following. Laterally compressed
body shape (like of sunfish, bluegill and angelfish) is good
for leisurely swimming and hiding or for predators who need
stability for attack and manoeuvrability.
III. BUOYANCY CONTROL
Underwater vehicles use buoyancy control mainly to
submerge and surface or compensate for a changing weight
For example, [6] uses a variable buoyancy control to
compensate the weight of the payload fetched from the
bottom. While buoyancy is traditionally controlled by
compressed air and water, alternative methods have also been
reported. For example, buoyancy control of [7] is inspired by
sperm whales and uses oil temperature regulation for decent
and ascent.
The novel aspect of this research is that it uses buoyancy
regulation not only to compensate its negative buoyancy and
for depth control but also to change the orientation of the
vehicle.
The general idea is to use two ballast tanks at both sides of
a compressed streamlined body (Fig. 3). In the horizontally
compressed mode both tanks are used to regulate buoyancy.
In the vertically compressed mode, only the upper tank is
filled with the air and is used to control vehicles buoyancy.
When the vehicle has zero buoyancy then its density is
equal to the density of sea-water. The density of the vehicle
depends on its mass and volume. Since the vehicle will be
constantly rebuilt as the tests proceed and requirements
change, buoyancy of the vehicle can calculated only
approximately. Even when the prototype is built ready, its
buoyancy still depends on the sensors added or removed, the
content of the water containers and the salt content of seawater as well as on the density of air left in the compressed
air balloon. Therefore the buoyancy is fine-tuned with the
help of the trimming weights.
Vehicle’s buoyancy is designed to be negative. To have
zero buoyancy the air balloons are initially filled with 0.5
litres of air. The estimated overall weight of the vehicle is
approximately 15 kg. The compressed air balloon can hold
64 litres of air and both air expansion chambers can contain
up to 1.5 litres of air. That means that total lift force can be
approximately 2.5 kg and for changing orientation we have
about 1 kg directional lift force available.
IV. MECHANICAL STRUCTURE
The layout of the interior of the vehicle is represented in Fig.
4. Since our aim is to keep the cost of the vehicle low, the
prototype is built from off-the shelf components that are
easily replaceable.
Fig. 3. The design concept of the orientation changing buoyancy
control.
The streamlined hull of the vehicle houses the pneumatic
system, servos and stepper motors to control the control
surfaces, electronic circuits, batteries and sensors. The
supporting rod of the vehicle is used to fix these components.
The hull is floodable and made of fibreglass. Openings in the
hull are for cameras, forward and bottom-looking sonar,
lights, and to attach the rudders to the body of the vehicle.
Variable ballast tanks are placed at both sides of the vehicle
and made of PVC tubes. Trimming weights can be added or
removed from the bow and stern ends of these tubes to
compensate changes in buoyancy when modules are changed
or payload is added.
The compressed air balloon is fixed between the supporting
rods between the ballast tanks at the centre of the vehicle.
The PVC tube in the bow part of the vehicle is a watertight
compartment sealed with a silicon sealant and fixed between
the supporting skeleton. This compartment houses control
electronics and batteries for emergency cases. When the offboard power supply is disconnected or communication with
the surface control unit is lost, the vehicle will surface.
The watertight PVC tube housing in stern of the vehicle is for
a stepper-motor powering the caudal fin.
Fig. 4. Internal layout of the vehicle
(1-fin, 2-stepper, 3-PVC tube, 4-compressed air tank, 5-rubber tank,
6-fin, 7-stepper, 8-camera, 9-batteries and electronics)
sources of light on the sides of the vehicle is that there is
usually lots of floating plankton and parts of underwater
macrovegetation. Since camera is attached in the middle of
the vehicle, the scene is illuminated from the sides and
reflection is not as intensive as in case of a single source of
light. In addition to the halogen lights, to arrays of LEDs are
used for spectral analysis of vegetation (see Fig. 6 and Fig.
7).
Fig. 5. Partially completed prototype of the vehicle without the
floodable outer shell.
The void between the PVC tubes and the compressed air
balloon coincides with the opening in the hull and is meant
for a down-looking camera and sonar. There is a symmetrical
opening on the upper part of the hull that permits adding an
additional camera. When the vehicle is operated in a
vertically compressed configuration, aquatic environment
can be inspected at both sides of the vehicle. The void in the
bow in front of the watertight compartment is prepared to
host the bow sonar.
The empty regions at both sides of the watertight
compartment are prepared for containers of water samples
and sensors that analyse them (like salinometer, microscope,
PH- sensor, etc.). It is planned that these sensors are
interchangeable and are used depending on the
environmental conditions and the mission.
The partially completed prototype is represented in Fig. 5.
The voids on both sides at the bow will be completed with
water containers and sensors to analyse water samples.
Fig. 6. LED light source with housing.
As it was explained in introduction, water in Baltic Sea is
quite turbid and therefore visibility is low. Two halogen
lamps in front of the elevators are used to increase visibility
or to cancel out reflections when the vehicle operates in very
shallow water in a direct sunlight. The reason of having two
Fig. 7. LEDs and control electronics.
V. PNEUMATIC SYSTEM OVERVIEW
Buoyancy of the vehicle is controlled by regulating the
volume ratio of water and air in ballast tanks. The tanks have
outlets at both ends so that water can flow in or out. In the
middle of the tank, there is an air expansion chamber made
from rubber. When the chamber is filled with air, water will
drain from the tank and the buoyancy of the vehicle
increases.
Both ends of the ballast tanks have also a place for trimming
weights that can be used to compensate for the change of
weight of the vehicle or adjust the centre of gravity when
modules are changed or added.
The compressed air balloon supplies air at a pressure of
approximately 8 at. This air is used to inflate the air
expansion chambers in the ballast tanks (Fig. 8). The
compressed air balloon is connected to the chambers with air
inlet hoses. A Y-branch divides the inlet path to two
branches. There are two pairs of valves at both branches. The
first pair of valves is outlet valves that can be controlled
independently. This permits inflating or deflating only one of
the air expansion chambers at time when the vehicle is
operating in a laterally compressed configuration. The
second pair of valves is the check valves that prevent water
from intruding into the pneumatic system.
The air outlet hoses are attached between the air expansion
chambers and openings at the frontal part of the vehicle. Like
inlet paths, the outlet paths have check and outlet valves.
Both branches are connected together with additional Ybranches, a hose and an outlet valve near the frontal end of
the pressured air balloon. This valve is opened when the
vehicle is operating in a horizontally compressed
configuration. This guarantees an equal pressure in both air
expansion chambers and therefore improves pitch stability
and controllability of the vehicle.
Fig. 8. Pneumatics
(1-pressure equalizing valve, 2-air inlet, 3-input valves, 4-output
valves, 5- rubber tanks, 6-compressed air tank)
VI. CONTROL
The block diagram of the vehicle’s electronics is provided in
Fig. 9.
The control of the vehicle is hierarchical. The surface control
unit is a laptop computer connected to the GPS receiver and
accessible for a human operator via a graphical user
interface. The surface unit receives and stores sensor data
that can be later analysed by environmental scientists. It also
receives acknowledgement and state signals from the
underwater unit and sends down high level commands.
The surface and underwater control units are connected via
an Ethernet link. The underwater control system is in turn
hierarchical, consisting of a Strong ARM 400 MHz processor
on the highest level, TI MSP processor on the middle layer
and PIC processors on the lowest layer.
The X-Scale architecture based Intel Strong ARM processor
is responsible for communication with the surface unit, for
passing up camera data, for processing the received high
level commands, high level planning and controlling the next
control level.
It is connected to external data storage and colour camera via
an USB link and to the compass and inclinometer via RS232.
Connection to the lower level MSP processor is also
established through RS232 serial interface.
Texas Instrument’s MSP430 microcontroller is an ultra-low
power 16-bit RISC mixed-signal processor. It receives high
level commands from Strong ARM, decomposes the tasks
and passes commands down to the next level
microprocessors responsible for executing the commands. It
also reports back to the Strong ARM processor about the
success or failure of the commands.
Fig.9. Block diagram of electronics and control.
The lowest level of control consists of an array of
Microchips’ PIC18F1320 16-bit 40 MHz processors. These
processors are responsible for steering the actuators and
receiving sensor data. They are connected to the MSP
processor via I2C interface.
The first PIC processor is responsible for driving the motors
of the rudders and vents. The next processor drives the valves
of the pneumatic system. The third processor receives data
from the bottom and front-looking sonar, analysis the data
and passes it forward to the MSP processor.
The fourth PIC processor is used to establish proper lightning
conditions. Our preliminary experiments have shown that
spectral analysis would facilitate recognition and
categorization of underwater macrovegetation. First, it could
help a human expert to distinguish between plants and
uncovered seabed in case of low visibility. Second, we aim at
developing an image processing method that automatically
recognizes vegetation from a video tape. We have therefore
built an array of LEDs with different wavelengths (Fig. 7).
The PIC processor under consideration is thus responsible for
steering the arrays of LEDs and switching the halogen lamps
in and out when spectral analysis is performed.
The next PIC processor steers the vents of the containers of
water samples.
Additional processors can be added to process data from
additional sensors.
VII. CONCLUSIONS AND FUTURE WORK
This paper describes the preliminary prototype of an
underwater vehicle. The vehicle is built for environmental
inspection in shallow waters of Baltic Sea. The design
concept is based on the environmental restrictions, human
factors and cost requirements.
The novel aspect of the vehicle’s design is the buoyancy
control that permits using the vehicle in two orientations,
horizontally and vertically compressed. The orientation can
be changed by controlling ballast tanks at both sides of the
vehicle.
At present, the construction of the vehicle is mostly complete
but underwater test are not yet done. The next phase of this
study is therefore a careful testing of the system and all
subsystems of the vehicle both is pool environment and in a
natural environment where currents and surging is present,
and the bottom is uneven.
This semiautonomous vehicle is designed to be used in two
modes, a towing mode behind a boat and in a remotely
operated mode. The towing mode permits covering long
distances at high speed compared to the human diver and
therefore considerably increases the efficiency of
environmental inspection. We also anticipate that the towing
mode is easier to implement that the remotely operated mode
since the vehicle has less degrees of freedom and needs
fewer control surfaces.
We therefore first aim at building a working prototype that
can be used in the towing mode by environmental scientists.
This involves in fist hand implementation of depth control of
the vehicle with help of rudders and front and bottom looking
sonar at the same time maintaining yaw and roll stability. At
present, the preliminary control model is ready but not tested
yet.
The remotely operated mode in a laterally compressed
orientation is a more complex task and definitely implies
thorough investigation of control models and vehicle’s
kinematics and dynamics.
Parallel to the vehicle’s design we also work at image
recognition algorithms for classification of underwater
vegetation.
V. ACKNOWLEDGEMENT
This research is supported by Estonian Scientific Foundation
grant No. 5613 and by Environmental Investment Centre.
VI. REFERENCES
[1] R.L. Wernli. “AUVs – A Technology Whose Time Has
Come.”, 2002. Proceedings of the 2002 International
Symposium on Underwater Technology, 16-19 April 2002,
pp. 309-314.
[2] G. Oliver, D. Avia, M. de Diego, A. Ortiz and J.
Proenza. “RAO: A Low-Cost AUV for Testing.” OCEANS
2000 MTS/IEEE Conference and Exhibition, 11-14 Sept.
2000, pp.397 – 401.
[3] S. Williams, G. Dissanayake, H. Durrant-Whyte.
“Towards Terrain-Aided Navigation for Underwater
Vehicles.” Advanced Robotics, Vol. 15, No. 5, pp. 533-549
(2001).
[4] N. Kato, T. Inaba, “Guidance and Control of Fish Robot
with Apparatus of Pectoral Fin Motion.” Proceedings. of the
2001 IEEE Int. Conf. on Robotics and Automation, Leuven,
Belgium, May 1998, pp. 446 – 451.
[5] M. Sfakiotakis, D.M. Lane, J.B.c. Davies, “Review of
Fish Swimming modes for Aquatic Locomotion.” IEEE
Journal of Oceanic Engineering, Vol. 24, No. 2, April 1999,
pp. 237-252.
[6] K.S. Wasserman, J.L. Mathieu, M.I. Wolf, A. Hathi, S.E.
Fried and A.K. Barker “Dynamic Buoyancy Control of an
ROV using a Variable Ballast Tank”, OCEANS 2003.
Proceedings, Vol. 5, 22-26 Sept. 2003 pp- SP2888 - SP2893.
[7] D. McFarland, I. Gilhespy, E. Honary, “DIVEBOT: A
Diving Robot with a Whale-like Buoyancy Mechanism.”
Robotica, Vol. 21, 2003, pp. 385 – 398.